Well, maybe not so easy...; HwNote04

[John B. Milton]

Another in a series, so far about adding a second hard disk drive the UNIXpc

Well, I got one UNIXpc upgraded. It seems the .5M machine I got from DDS is
older than the one I'm working on now. Now that I look at the schematics, I
find that the older machine (the one I got working with two hard drives) must
not match the schematics. If it did, it wouldn't have worked. The machine I'm
working on now does seem to match the schematics, and is not working, just like
I would expect. Once I get ahold of the data sheet for the WD1010, I should
be able to figure out how to get this one working.

I am now very much in favor of a small daughter board for this upgrade. I
managed to fry the 26LS32 receiver chip on the second machine. This daughter
board would fit into the socket at 14M. The PAL in that socket would be put
on the duaghter board. For all of you who have or would be will to open up
your UNIXpc, please check this location (14M), and see if the chip is indeed
socketed. I need to know as soon as possible if anyone has this PAL soldered in.
Convergent seems to be good about socketing all the expensive chips and the PALs
on all the boards I've seen so far.

So far, the daughter board will have: 26LS31, 26LS32, 74LS175, the PAL from
14M, a 20 pin DIL male connector for the data cable to the second hard drive,
a 100 ohm resistor, four bypass caps. Another reason for the daughter board is
a firm mounting place for the connector.

Gary Sanders (gws at n8emr) had a neat idea which I think could be implemented
on the daughter board. There is at least one problem with the current release
of UNIX on the UNIXpc. That is the "slow down and lock up bug". With this bug,
the system gets slower and slower and finally locks up. It could be some
program on the clock tick callout that eventually takes up more the 1/60th of
a second. My guess for culprits there would be the window driver. The symptom
after it has already locked up is: The mouse pointer still moves around just
fine. The time in the status manager's window is the time it locked up. Typing
a key will cause the flashing of the cursor to skip a beat, but no characters
will get echoed.

Gary's idea is to have a long duration watch dog timer on the reset line.
The idea is to set the timer for, say 20 minutes. A program "petdog" would be
run out of cron on 5 minute intervals. This program would attempt so
operations such as disk or window, then reset the watchdog timer. As long as
the watchdog keeps being reset, the system stays up. If the watchdog does not
get reset, the system gets reset. Unfortunately, the XRST line on the expansion
bus is driven by a gate, so an expansion card can not reset the system. This
means the line must be driven from inside the machine.

Gary Sanders has the board I already got working. He will soon be stress
testing a setup with two drives. He has the second drive exteral in a PC box.
He also tells me some interesting news that Western Digital has a new hard
disk controller chip that does RLL and is almost pin compatable with the WD1010.
He'll post the details when he gets the data sheet.

Jan Isley (jan at bagend) says he will be looking into adding a 1.2M floppy soon.
Right now he's having problems getting postings out onto the net. The 1.2M
floppy looks like it might be possible because the Western Digital Floppy Disk
Controller has a pin on it to switch from 5.25" to 8" disk drives. The high
density floppy drives look just like 8" drives I'm told. There is also no
problem with changeing the pin under software control. I have an S-100 Z-80
system that has another FDC from the 279x series, the 2793. I have used the
S-100 with both 5.25" and 8" at the same time. Of course having two floppies
would at least require a new gd driver.

One more note while I'm thinking about it. As most of all of you know, if
you enter "s4test" or "S4TEST" to the 7 choice menu in the diagnostics, you
will be put in "expert" mode. Mostly, this mode allows you to run just one
test, or run it many times. The ? command at the "expert>" prompt displays
a full screen of help. The list of commands has the hard disk test routines
listed as #6. When you enter 6 or 6,0 it will run a default list of tests.
When you enter the command as 6,n you can run a specific sub-test. The know
subtests are: 1:recal, 2:format, 4:read sectors across disk, 5: random seek
and read id, 6: non-destructive read, 8: spare a sector, 12: print VHB and BBT,
16: go to interactive device test mode ("i>" prompt, exit with "q"), 23:
DESTRUCTIVE surface test. DO NOT RUN THIS EXCEPT ON NEW DISKS!!, 24: park head.
The part they don't tell you here is that all these tests can be run on the
second hard drive by using the test number 2 instead of 6. So, 2,12 will
display information on the second hard drive. Note that the diags also check
to see if you mother board is at least rev P5.1 before it will access the
second hard drive. I can't think of any way to fudge the revlev in the diags.
One would have to swap the second drive to the first for formatting and testing

BTW, the diagnostics are bootable just like /unix. The diagnostic disk is a
mountable file system. If you copy the s4diag file from the diagnostic floppy
onto you hard disk, you can boot the system with diags without using the floppy
drive. The trick to doing this is to get the verbose loader onto the hard drive.
The way to do this is to format a floppy with the fdfmt.vl command, then copy
the verbose loader from the floppy to the hard disk with the /etc/ldrcpy program

The next time the system is booted, it will stop and ask which drive you want
to boot from, then which file you want to load from. It needs the beginning /
for the file name, e.g. "/unix". There is one very obvious problem with using
the verbose loader, of course. If you system resets from a power failure, it
will not reboot, it will wait asking questions. This may be a feature at some
installations.

Sorry for this one being so boring.

John
-- 
John Bly Milton IV, jbm at uncle.UUCP, n8emr!uncle!jbm at osu-cis.cis.ohio-state.edu
home (614) 294-4823, work (614) 764-4272;  Send vi tricks, I'm making a manual